With the explosion of the Internet, organizations such as businesses, non-profit foundations, schools, and the like, are fast making reliable connections to the Internet an integral part of their networking strategy. However, in order to reduce the chance of a potential catastrophic shutdown, organizations are establishing multiple points of connection to the Internet. Organizations that desire the additional reliability often establish multiple connections through several different Internet Service Providers (ISP(s)). Each ISP then assigns an IP address or range of IP addresses to the organization for Internet access. Network border routers, using network address translators (NAT), are then employed by the organization to manage their Internet access using the ISP assigned addresses.
In addition to the potential for increased reliability, maintaining connections through multiple ISP(s) enables the organization to perform load-balancing by lowering the number of internal organization nodes (or devices) linking to the Internet through a single link. By distributing the load through multiple links the organization may optimize performance and significantly reduce wait times.
While reliability and performance considerations have contributed to increased deployment of multiple Internet connections, such implementations are not without their perils. For example, in the ideal situation, when the connectivity through one ISP fails, connectivity through the other ISP(s) is expected to enable the organization to preserve its connectivity to the Internet, and thereby reduce potential impact to the organization. However, a problem can occur when one ISP fails.
When one ISP fails, outbound traffic must be rerouted to take advantage of another available ISP connection. However, there is no way to control the ISP used for returning network traffic. Because the ISP used for the return traffic is not selectable by the organization, there is a high likelihood that responses to outbound traffic will result in incomplete transactions for the organization, and potentially lost profits. The inability to manage the return network traffic route will also result in an inability for the organization to balance its incoming Internet traffic.